The term "high frequency" is used herein to refer to operating frequencies at which electromagnetic radiation and electromagnetic interference between proximate circuits becomes a problem. As an example, many microwave circuits are configured to operate at about 30 gHz.., and that frequency represents a useful application of the present invention. In many of its aspects, the invention can be considered to be broadly useful in the high frequency range from about 800 mHz. to 100 gHz..
In many cases, in such high frequency devices, it is necessary to isolate a given integrated circuit (or group of circuits) from others so that the circuits will not interfere with each other. The electromagnetic radiation which passes from one circuit to the other in the form of electromagnetic interference, can serve as an unwanted feedback path. One of the results can be the creation of unwanted oscillations in the circuit. Another can be the alteration of device characteristics from a desired characteristic intended by the designer to an entirely different characteristic for the circuit operating with the unanticipated feedback path. The nature of the feedback path is very affected by the operating frequency of the device, the nature of the coupling, the presence of adjacent conductors, improperly designed shielding devices, and the like. In some cases, such as when the size of the cavities in which the integrated circuits are mounted is greater than one-half wavelength, unwanted resonances can be generated near the working frequency due to cavity resonance.
FIG. 6 illustrates in perspective a conventional semiconductor device, sometimes known as a modular structure, designed to alleviate some of the foregoing problems. Integrated circuits 4 and 6 are illustrated as two circuit elements which are intended to be isolated. The circuits are electrically interconnected by a conductor 5. In the illustrated embodiment, each of the circuits 4, 6 is formed on its own substrate. Even if the circuits were originally formed on the same wafer, in order to be used in the configuration of FIG. 6, they are required to be diced or otherwise separated for individual mounting.
In order to provide shielding between the integrated circuits 4, 6 a shielding circuit mount 8 is provided having a pair of upstanding walls 7 interposed between the circuits 4, 6 and rising a substantial distance above the plane 9 of the substrate 8 on which the integrated circuits are mounted. The substrate 8 is formed of conductive metal, typically copper or copper-plated tungsten, and is adapted to provide a ground plane under the circuits 4, 6 and intermediate shields 7 separating those circuits, but leaving a narrow passage for the conductor 5. The integrated circuit chips 4, 6 as well as the conductor 5 are bonded to the substrate 8 as by solder, a conductive adhesive material or the like.
The walls 7 typically range in height from about 0.01 to about 0.1 wavelengths of the operating frequency of the circuit. With that dimension of height, and maintaining the channel between the walls 7 narrow but adequate for passage of the conductor 5, substantial shielding is provided between the circuits 4, 6. However, that shielding is achieved at the cost of assembly procedures, and an increase in size of the completed device. With respect to size, it is clear that the mounting substrate must be large and strong enough to tolerate the manipulative assembly techniques, and that the walls 7 must be sufficiently thick to be self-supporting.
More particularly, in order to use the shielding substrate 8 as shown in FIG. 6, the integrated circuits 4, 6 must be formed on separate substrates, must be assembled onto the substrate, and the conductor 5 must also be arranged on the substrate and interconnected to the respective semiconductors in order to provide a completed assembly. Since a metallic conductive body 8 is used it is difficult to achieve the desired size and weight reduction for the semiconductor device, and the aforementioned assembly procedures also serve to increase cost.
FIG. 7 illustrates another prior art approach as described in Japanese published patent specification No. 63-143856. FIG. 7 shows a plurality of serially connected distributed amplifiers 14a, 14b, 14c, each having a plurality of active elements arranged in multiple rows on a planar surface 15a of a semi-insulating substrate 15. An input terminal 16 feeds the first group of active elements 14c which then drive in cascade the remaining groups 14b, 14a to drive an output terminal 13. Thus, a microwave signal interposed on input terminal 16 is amplified and reproduced at output terminal 13.
In order to provide a measure of shielding between the groups of amplifiers 14a, 14b, 14c, grounded wiring patterns 17 are formed on the surface 15a of the substrate and grounded by means of vias 18. The vias 18 penetrate the substrate 15 and are filled with conductive metal, forming a connection to a ground plane to maintain the surface deposited conductors 17 at a ground potential. However, the thickness of the conductors 17 is typically only about 2 microns, and therefore such conductors are not highly effective in performing a shielding function for electromagnetic radiation. While the surface mounted shields 17 might be effective for interference which would otherwise propagate along the surface of the substrate, the height of the shields limits any substantial effect on the very significant EMI radiated in the air paths over the shields. Such radiation, in effect, provides an unwanted feedback path between the groups of amplifiers 14a, 14b, 14c.
The effect of that feedback path can be significant as will now be shown. It is known that electromagnetic energy of microwave frequency which propagates from a driving circuit to another circuit travels not only along the surface on which the circuits are mounted, but also in the space surrounding the circuits. It is believed that about 80% of the energy is propagated in the space other than on the substrate surface. Therefore, it will be appreciated that while surface mounted conductors 17 are adequate to block surface traveling interference, the majority of the energy, i.e., that propagated in the space surrounding the circuit, remains substantially unaffected. The limitations of this shielding technique will therefore be apparent.